Association Between County-Level Change in Economic Prosperity and Change in Cardiovascular Mortality Among Middle-aged US Adults
Sameed Ahmed M. Khatana, MD, MPH; Atheendar S. Venkataramani, MD, PhD; Ashwin S. Nathan, MD;
Elias J. Dayoub, MD, MPP; Lauren A. Eberly, MD, MPH; Dhruv S. Kazi, MD, MSc, MS; Robert W. Yeh, MD, MSc;
Nandita Mitra, PhD; S. V. Subramanian, PhD; Peter W. Groeneveld, MD, MS
IMPORTANCEAfter a decline in cardiovascular mortality for nonelderly US adults, recent stagnation has occurred alongside rising income inequality. Whether this is associated with underlying economic trends is unclear.
OBJECTIVETo assess the association between changes in economic prosperity and trends in cardiovascular mortality in middle-aged US adults.
DESIGN, SETTING, AND PARTICIPANTSRetrospective analysis of the association between change in 7 markers of economic prosperity in 3123 US counties and county-level
cardiovascular mortality among 40- to 64-year-old adults (102 660 852 individuals in 2010).
EXPOSURESMean rank for change in 7 markers of economic prosperity between 2 time periods (baseline: 2007-2011 and follow-up: 2012-2016). A higher mean rank indicates a greater relative increase or lower relative decrease in prosperity (range, 5 to 92;
mean [SD], 50 [14]).
MAIN OUTCOMES AND MEASURESMean annual percentage change (APC) in age-adjusted cardiovascular mortality rates. Generalized linear mixed-effects models were used to estimate the additional APC associated with a change in prosperity.
RESULTSAmong 102 660 852 residents aged 40 to 64 years living in these counties in 2010 (51% women), 979 228 cardiovascular deaths occurred between 2010 and 2017.
Age-adjusted cardiovascular mortality rates did not change significantly between 2010 and 2017 in counties in the lowest tertile for change in economic prosperity (mean [SD], 114.1 [47.9] to 116.1 [52.7] deaths per 100 000 individuals; APC, 0.2% [95% CI, −0.3% to 0.7%]).
Mortality decreased significantly in the intermediate tertile (mean [SD], 104.7 [38.8] to 101.9 [41.5] deaths per 100 000 individuals; APC, −0.4% [95% CI, −0.8% to −0.1%]) and highest tertile for change in prosperity (100.0 [37.9] to 95.1 [39.1] deaths per 100 000 individuals;
APC, −0.5% [95% CI, −0.9% to −0.1%]). After accounting for baseline prosperity and demographic and health care–related variables, a 10-point higher mean rank for change in economic prosperity was associated with 0.4% (95% CI, 0.2% to 0.6%) additional decrease in mortality per year.
CONCLUSIONS AND RELEVANCEIn this retrospective study of US county-level mortality data from 2010 to 2017, a relative increase in county-level economic prosperity was significantly associated with a small relative decrease in cardiovascular mortality among middle-aged adults. Individual-level inferences are limited by the ecological nature of the study.
JAMA. 2021;325(5):445-453. doi:10.1001/jama.2020.26141
Editorialpage 439 Supplemental content CME Quiz at
jamacmelookup.comandCME Questionspage 486
Author Affiliations: Author affiliations are listed at the end of this article.
Corresponding Author: Sameed Ahmed M. Khatana, MD, MPH, Division of Cardiovascular Medicine, Perelman School of Medicine, University of Pennsylvania, 3400 Civic Center Blvd, Philadelphia, PA 19104-5162 (sameed.khatana@
pennmedicine.upenn.edu).
JAMA | Original Investigation
C
ardiovascular mortality rates for nonelderly adults in the US over the last decade have stopped declining and, for some populations, have even increased.1-3 The factors driving these trends in cardiovascular mortality are poorly understood but may be related to underlying eco- nomic trends given the strong association between cardio- vascular disease and economic indicators such as income.4Understanding how economic trends have influenced cardiovascular mortality is crucial for devising strategies to address this stagnation. This is especially true given rising levels of income inequality within the US.5The eco- nomic recovery after the 2008-2009 recession has been uneven, leading to significant disparities in economic pros- perity between different areas of the US. Several studies have demonstrated the association between individual- level social determinants of health, as well as markers of community-level economic activity and cardiovascular health and mortality.6-8However, due to the cross-sectional design of many of these studies, it is unclear whether changes in economic prosperity correspond to changes in health outcomes and whether stagnating mortality rates could be associated, in part, with worsening community economic prosperity.
This study evaluated whether changes in relative eco- nomic prosperity in the postrecession period have been asso- ciated with trends in cardiovascular mortality rates for middle- aged adults—the segment of the population in which the departure from long-standing secular declines in cardiovas- cular mortality has been previously noted—using county- level mortality and economic data.3
Methods
This analysis was considered exempt from review by the University of Pennsylvania institutional review board guide- lines because it uses publicly available data routinely col- lected for public health purposes.
Mortality Data
We obtained restricted mortality data from the National Cen- ter for Health Statistics, which includes details on every re- corded death in the US including age, sex, year of death, race/
ethnicity, cause of death, and county of residence at the time of death from January 2010 to December 2017.9
Change in Economic Prosperity
Change in economic prosperity was based on the Distressed Communities Index, which is composed of 7 markers of economic activity drawn from the US Census Bureau’s American Community Survey 5-Year Estimates and Business Patterns data sets and is available for all counties with more than 500 residents.10We calculated the absolute change in the following markers of economic prosperity between the baseline (2007-2011) and follow-up (2012- 2016) periods: (1) housing occupancy rate, (2) ratio of the county median household income to state median house- hold income, (3) percentage of 25- to 64-year-old adults
working, (4) percentage of the adult population with a high school education, (5) percentage of the population with income above the poverty threshold, (6) percentage change in the number of business establishments between the first and last years of the time period, and (7) percentage change in the number of jobs between the first and last years of the time period. Change in economic prosperity was then deter- mined by ranking counties for change in each of these mark- ers on a scale from 0 to 100 and calculating an unweighted mean of these ranks. Counties with a higher mean rank had a greater increase, or lower decrease, in economic prosper- ity relative to counties with a lower mean rank. Change in economic prosperity follows an approximately normal dis- tribution (eFigure 1 in theSupplement). Baseline (2007- 2011) economic prosperity levels were determined by taking the mean of the rankings of the baseline value of each eco- nomic prosperity marker on a scale of 0 to 100, with a higher rank indicating greater relative baseline economic prosperity compared with a lower-ranked county. Tertiles of baseline economic prosperity levels across the US are dis- played in eFigure 2 in theSupplement.
Data sources for other demographic and health care–
related covariates included are listed in eMethods 1 in the Supplement.
Outcomes
The primary outcome measure was the annual percentage change (APC) in county-level, aggregated, age-adjusted (to the 2000 US population) cardiovascular mortality rates per 100 000 individuals for 40- to 64-year-old adults. Distribu- tions for baseline and change in mortality are displayed in eFigure 3 in theSupplementand yearly mortality levels across the US by tertiles are displayed in eFigures 4, 5, 6, 7, 8, 9, 10, and 11 in theSupplement.
Secondary outcomes included mortality from cardio- vascular disease subgroups (ischemic heart disease and stroke), all causes, diseases of the circulatory system (a broader definition of cardiovascular disease), and cancer.
Cause of death was based on International Statistical Classi- fication of Diseases and Related Health Problems, Tenth Revi- sion codes and utilized previously used classifications (eTable 1 in theSupplement).11
Key Points
QuestionAre changes in county-level economic prosperity associated with changes in cardiovascular mortality among middle-aged adults (aged 40-64 years) in the US?
FindingsIn this retrospective analysis of county-level mortality data from 3123 US counties from 2010 to 2017, every 10-point greater change in economic prosperity from baseline to follow-up (range, 5 to 92) was significantly associated with a 0.4% lower cardiovascular mortality rate per year among middle-aged adults.
MeaningIn US counties from 2010 to 2017, a relative increase in economic prosperity was associated with a small relative decrease in cardiovascular mortality among middle-aged adults.
Statistical Analysis
Counties were divided into tertiles based on change in pros- perity. Annual, population-weighted, mean mortality rates were calculated for each tertile. Mean APC in mortality rates from 2010 to 2017 was estimated for each tertile using the gen- eralized linear mixed model (GLMM), described here, with a linear year predictor.
To estimate the additional APC associated with a relative change in economic prosperity, we used GLMMs, which allow for the analysis of longitudinal and hierarchical data with nonnormal distributions. A negative binomial distribution with a log link was used (eMethods 2 in theSupplement). The primary model included baseline economic prosperity, year, and an interaction between year and change in prosperity (on a continuous scale). The interaction term was the primary variable of interest with an interpretation as the estimated difference in the APC between 2 counties with a difference of 10 ranks for mean rank for change in economic prosperity.
The model included county-level time-varying demographic and health care–related variables: proportion of county resi- dents that are female, non-Hispanic Black, Hispanic, have diabetes, have obesity, and have health insurance, and den- sity of hospital beds and primary care physicians and 1 time- invariant variable: proportion of residents living in rural areas in 2010. A spatial power covariance structure accounted for the longitudinal nature of the data and state random inter- cepts for clustering of counties within states. Random effects were assumed to have a normal distribution with a mean of zero. Robust standard errors were used.
Secondary analyses included examining secondary out- comes as listed above. Given significant differences in cardio- vascular mortality between non-Hispanic White and racial and ethnic minority populations, as well as between men and women previously noted,3we analyzed race and ethnicity subgroups and male and female mortality rates separately.
Racial and ethnic groups with small numbers of residents in many counties were not included in subgroup analyses due to statistical instability. Deaths were assigned to sex and race/
ethnicity subgroups based on reported values on death cer- tificates, which have been shown to adequately classify Black and White race as well as Hispanic ethnicity (greater than 90% agreement between death certificates and self-reported race and ethnicity).12Mortality rates for other age groups were also analyzed.
For each model, we calculated the absolute additional annual change in mortality rates associated with a change in prosperity (based on the population-weighted median mortality rate in 2010). We also constructed GLMMs with alternative distributions, with change in economic prosperity as a quadratic term and as a categorical variable to assess whether our findings were robust to the model assumptions.
We also constructed alternative conditional GLMMs with- out state random effects, which allowed comparison of goodness-of-fit statistics of different models (which was not possible in the primary analysis). The primary GLMM was also estimated stratified by different levels of baseline eco- nomic prosperity, and with change in each of the prosperity markers separately.
All analyses were weighted by the relevant county population, unless specified otherwise. Data are presented as means with SDs or 95% CIs or medians and interquartile ranges (IQRs). All P values were 2-sided and values of less than or equal to .05 were considered statistically signifi- cant. Because of the potential for type I error due to mul- tiple comparisons, findings for secondary analyses and secondary end points should be interpreted as exploratory.
All statistical analyses were conducted using SAS version 9.4 (SAS Institute).
Results
A total of 3123 US counties were included (Figure 1), with 20 counties excluded that had economic prosperity markers un- available. Data on covariates were available for all included counties. Mean rank for change in economic prosperity ranged from 5.4 to 91.9 overall (mean [SD], 49.9 [13.9]), from 5.4 to 43.8 in the lowest tertile, 43.8 to 56.0 in the intermediate ter- tile, and 56.1 to 91.9 in the highest tertile for change in eco- nomic prosperity. In 2010, a total of 102 660 852 individuals aged 40 to 64 years lived in the included counties (51.0% wom- en). Demographic and health care–related variables, in 2010, for the 3 groups of counties are listed in Table 1. Counties in the lowest tertile for change in prosperity had the lowest me- dian population (7078; IQR, 3172-16 684), highest percentage of residents in rural areas (25.9%; IQR, 10.9%-53.3%), highest percentage of residents with diabetes (9.1%; IQR, 8.0%- 10.4%) and obesity (29.3%; IQR, 26.9%-32.4%), lowest me- dian number of primary care physicians per 100 000 resi- dents (68.5; IQR, 47.2-88.7), and the lowest percentage of county residents with health insurance (85.8%; IQR, 81.3%- 89.9%). The mean prevalence of diabetes and obesity in- creased significantly in all 3 groups of counties from 2010 to 2016 (eTable 2 in theSupplement). The total number of 40- to 64-year-old residents decreased in counties in the lowest tertile for change in prosperity, but increased in the other 2 groups of counties over the study period (eTable 3 in the Supplement). County characteristics by baseline prosperity ter- tiles are listed in eTable 4 in theSupplement.
Between the baseline and follow-up periods, counties in the lowest tertile for change in economic prosperity experi- enced a median decrease in 4 of 7 markers of prosperity:
housing occupancy rate, ratio of county median household income to state median household income, percentage of 25- to 64-year-old adults working, and percentage of the popula- tion with income above the poverty threshold (Table 2).
These counties experienced a median increase in the per- centage of adults with a high school education. The median percentage change in business establishments over a period increased from −4.9% (IQR, −7.8% to −1.4%) in the baseline period to −1.7% (IQR, −5.0% to 1.1%) in the follow-up period.
The median percentage change in employment over each period increased from −3.7% (IQR, −8.5% to 1.4%) to 0%
(IQR, −6.1% to 4.7%).
Counties in the intermediate tertile experienced a me- dian decrease in 3 markers: housing occupancy rate, percentage
of 25- to 64-year-old adults working, and percentage of the population with income above the poverty threshold. There was no change in the median ratio of county median house- hold income to state median household income. The median percentage change in business establishments over a period increased from −6.5% (IQR, −9.1% to −3.8%) to 0.6% (IQR,
−2.4% to 3.6%). The median percentage change in employ- ment over each period increased from −6.3% (IQR, −11.1% to
−1.7%) to 4.2% (IQR, −0.8% to 9.2%).
Counties in the highest tertile experienced a median in- crease in all markers of economic prosperity. The median per- centage change in business establishments over a period in- creased from −7.4% (IQR, −10.9% to −4.2%) to 2.0% (IQR, −1.1%
to 6.3%). The median percentage change in employment over each period increased from −8.7% (IQR, −14.0% to −3.6%) to 8.1% (IQR, 2.3% to 13.6%).
The mean (SD) (population-unweighted) baseline pros- perity levels were 53.3 (28.0) in the lowest tertile, 52.5 (28.4) in the intermediate tertile, and 44.2 (29.4) in the highest ter- tile for change in prosperity.
Primary Analysis
There were 979 228 cardiovascular deaths among 40- to 64- year-old adults between 2010 and 2017. Age-adjusted cardio- vascular mortality rates did not change significantly from 2010 to 2017 in counties in the lowest tertile for change in prosper-
ity (mean [SD], 114.1 [47.9] to 116.1 [52.7] deaths per 100 000 individuals; mean APC, 0.2% [95% CI, −0.3% to 0.7%]). Mor- tality rates decreased significantly in the intermediate tertile (mean [SD], 104.7 [38.8] to 101.9 [41.5] deaths per 100 000 in- dividuals; mean APC, −0.4% [95% CI, −0.8% to −0.1%]) and highest tertile for change in prosperity (mean [SD], 100.0 [37.9]
to 95.1 [39.1] deaths per 100 000 individuals; mean APC, −0.5%
[95% CI, −0.9% to −0.1%]) (Figure 2; eTable 5 in theSupple- ment). After accounting for baseline economic prosperity and time-varying demographic and health care–related factors, for every 10-point higher mean rank for change in economic pros- perity, counties had an additional 0.40% (95% CI, 0.22% to 0.58%) decrease in mortality per year (Table 3). Distribution of random effects, sensitivity tests, and goodness-of-fit sta- tistics in the conditional GLMMs suggest that the main model was appropriately specified (eFigure 12, eTable 6, and eTable 7 in theSupplement).
Secondary Analyses
Ischemic heart disease mortality rates decreased signifi- cantly in all 3 groups of counties (mean APC, −0.7% [95% CI,
−1.2% to −0.2%] for counties in the lowest tertile, −1.3%
[95%, −1.7% to −0.9%] for counties in the intermediate ter- tile, and −1.7% [95% CI, −2.3% to −1.1%] for counties in the high- est tertile for change in prosperity) (Figure 2; eTable 8 in the Supplement). Stroke mortality rates did not change signifi- Figure 1. Counties by Tertile of Change in Economic Prosperity and Change in Age-Adjusted Cardiovascular Mortality Rates
(40- to 64-Year-Old Adults)
Change in economic prosperity from baseline (2007-2011) to follow-up (2012-2016) periods
A
Lowest tertile Intermediate tertile Highest tertile Not included
Absolute change in age-adjusted cardiovascular mortality rates for 40- to 64-year-old adults from 2010 to 2017
B
Lowest tertile Intermediate tertile Highest tertile Not included
Maps are based on the US National Atlas Equal Area Projection and reflect county geographic size and not population. A, Change in economic prosperity is the unweighted mean of the ranks for change in the 7 markers of economic prosperity between baseline (2007-2011) and follow-up (2012-2016). A higher mean rank indicates a greater relative increase or lower relative decrease in economic prosperity compared with other counties. The mean rank for change in economic prosperity ranged from 5.4 to 43.8 for the lowest tertile (n = 1041), 43.8 to 56.0 for the intermediate tertile (n = 1041), and 56.1 to 91.9 for the highest tertile (n = 1041); 20 counties did not have data available and were not included. The lightest hue indicates relative improvement in prosperity.
B, The change in age-adjusted cardiovascular mortality was −611 to −14 deaths per 100 000 individuals in the lowest tertile (n = 1041), −14 to 21 in the intermediate tertile (n = 1041), and 21 to 666 in the highest tertile (n = 1041);
20 counties did not have data available and were not included. The lightest hue indicates a declining or stable cardiovascular mortality rate.
The primary analysis in Table 3 accounts for changes in mortality rates across all years from 2010 to 2017. Distribution of baseline economic prosperity levels and annual cardiovascular mortality rates across the US are available in eFigures 2, 4, 5, 6, 7, 8, 9, 10, and 11 in theSupplement.
cantly in the 3 tertiles. Change in economic prosperity was as- sociated with change in ischemic heart disease (−0.51% [95%
CI, −0.74% to −0.29%]) and stroke (−0.26% [95% CI, −0.47%
to −0.04%]) mortality rates (Table 3). All-cause mortality rates increased in counties in the lowest and intermediate tertiles (mean APC, 0.9% [95% CI, 0.7% to 1.2%] and 0.6% [95% CI, 0.3% to 0.9%], respectively) but did not change significantly in the highest tertile (mean APC, 0.01% [95% CI, −0.3% to 0.3%]) (Figure 2; eTable 9 in theSupplement). Change in eco- nomic prosperity was significantly associated with change in all-cause mortality rates (−0.47% [95% CI, −0.58% to −0.35%]) (Table 3).
Trends in cardiovascular mortality rates in each tertile for change in economic prosperity by race and ethnicity subgroups as well as for men and women are displayed in eTables 10 and 11 in theSupplement. The association between change in economic prosperity and additional APC in cardiovascular mortality rates was significant for each race and ethnicity subgroup (non-Hispanic Black: −0.60%
[95% CI, −0.92% to −0.27%], non-Hispanic White: −0.53%
[95% CI, −0.72% to −0.33%], and Hispanic [all races]:
−0.57% [95% CI, −1.05% to −0.08%]) and for women
(−0.58% [95% CI, −0.88% to −0.27%]) and men (−0.35%
[95% CI, −0.57% to −0.12%]) (Table 3). Mortality rates for other age groups (20-39 years of age and 65 years and older), cancer-specific mortality, and mortality from disease of the circulatory system are listed in eTables 12 and 13 in theSupplement. The association between change in eco- nomic prosperity and change in cardiovascular mortality rates was statistically significant in each tertile of baseline prosperity levels (eTable 14 in theSupplement). Increase in each of the individual economic prosperity markers, except the proportion of adults with a high school education, was associated with a significant decrease in the APC for cardio- vascular mortality (eTable 15 in theSupplement).
Discussion
From 2010 to 2017, US counties with the lowest relative im- provements in economic prosperity experienced no change in cardiovascular mortality rates for middle-aged adults. Dur- ing this period, cardiovascular mortality rates declined sig- nificantly in the remainder of US counties. A relative increase Table 1. County Characteristics (2010) by Tertile of Change in Economic Prosperitya
Characteristic
Median (IQR)
Lowest tertile Intermediate tertile Highest tertile
No. of countiesb 1041 1041 1041
Demographics
County population (No. of individuals aged 40-64 y)c 7078 (3172-16 684) 11 944 (5051-31 658) 8964 (4052-23 813)
Percentage of residents living in rural areasd 25.9 (10.9-53.3) 8.6 (1.6-27.3) 4.5 (1.1-20.3)
Percentage of residents aged 40-64 ye
Women 50.9 (50.0-51.6) 51.2 (50.7-51.9) 51.2 (50.5-51.7)
Men 49.1 (48.4-50.0) 48.8 (48.1-49.3) 48.8 (48.3-49.5)
Hispanic (all races) 3.3 (1.6-9.7) 5.2 (2.4-12.9) 9.1 (2.8-20.8)
Non-Hispanic Black 6.2 (1.5-13.9) 8.8 (2.9-15.7) 7.4 (2.6-18.6)
Non-Hispanic White 82.1 (67.3-92.2) 75.0 (55.5-88.6) 68.9 (49.9-83.0)
Cardiovascular risk factors Percentage of residents aged ≥20 yf
Diabetes 9.1 (8.0-10.4) 8.9 (7.8-10) 8.5 (7.6-9.5)
Obesity 29.3 (26.9-32.4) 27.9 (24.3-30.7) 26 (22.9-29.5)
Health care–related variables
Primary care physicians per 100 000 residentsg 68.5 (47.2-88.7) 70.8 (55.7-88.1) 73.6 (55.6-90.2)
Hospital beds per 100 000 residentsh 314.2 (191.0-445.5) 268.6 (195.1-390.7) 256.8 (166.3-356.7)
Health insurance coverage, percentage of residents aged 40-64 yi
85.8 (81.3-89.9) 85.2 (80.7-89.4) 83.4 (79.9-86.7)
Abbreviation: IQR, interquartile range.
aChange in economic prosperity is the unweighted mean of the ranks for change in the 7 markers of economic prosperity between baseline (2007-2011) and follow-up (2012-2016). A higher mean rank indicates a greater relative increase or lower relative decrease in economic prosperity compared with lower-ranked counties. The mean rank for change in economic prosperity ranged from 5.4 to 43.8 for the lowest tertile, 43.8 to 56.0 for the intermediate tertile, and 56.1 to 91.9 for the highest tertile.
bIncludes all US counties except for 20 counties that did not have economic prosperity markers available.
cUnweighted calculation. All other calculations weighted by county population.
dRural areas are defined by the US Census Bureau as areas with fewer than 2500 residents.
eSex, race, and ethnicity as reported on death certifications. Racial/ethnic groups other than Hispanic (all races), non-Hispanic Black, or non-Hispanic White were not included in analyses due to the small number across most US counties.
fDiabetes and obesity prevalence are based on self-reported diagnosis, height, and weight from the Centers for Disease Control and Prevention’s Behavioral Risk Factor Surveillance System.
gThe number of primary care physicians per county was obtained from the Health Resources & Services Administration Area Health Resource File.
hThe number of hospitals beds per county was obtained from the American Hospital Association annual survey.
iProportion of county residents with health insurance was obtained from the US Census Bureau’s Small Area Health Insurance Estimates program.
in county-level economic prosperity was associated with a rela- tive decrease in annual cardiovascular mortality rates for middle-aged adults in the US during this period.
There may be multiple mechanisms by which economic factors influence health. Job insecurity and income volatility may be associated with cardiovascular events.4,13,14Worsen- ing economic prosperity may lessen social cohesion and in- crease income inequality, contributing to a community’s health.15The postrecession period in the US has seen the rise of “deaths of despair” (eg, deaths from drug poisoning, sui- cide, alcoholic liver disease), which have also been attributed to worsening social cohesion.16This study suggests that in cer- tain areas of the country, a relative decrease or stagnation in economic prosperity was associated with mortality from causes beyond those typically considered as deaths of despair. Pos- sible biological mechanisms for how economic stress can worsen cardiovascular health include upregulation of inflammation.17It is also possible that worsening economic prosperity may lead healthy individuals to emigrate, result-
ing in higher mortality rates for the remaining population.
Counties in the lowest tertile for change in economic prosper- ity had a declining population; however, the analysis ac- counted for changes in age structure as well as changes in the prevalence of 2 important cardiovascular risk factors: diabe- tes and obesity.
A larger rural population was noted in counties with a decrease or stagnation of economic prosperity, in line with prior studies that have described an increase in cardiovascu- lar mortality rates in the past decade in rural areas.2,18,19 Rural areas of the US experienced a disproportionately greater increase in unemployment during the 2008-2009 economic recession and have not experienced the same level of recovery as other parts of the country.20This analysis also found that the association between economic prosperity and cardiovascular mortality was significant for non-Hispanic Black and Hispanic individuals, along with non-Hispanic White individuals. Studies of deaths of despair have high- lighted the association between economic factors and Table 2. Economic Prosperity Markers by Tertile of Change in Economic Prosperity (Values Not Weighted by County Population)a
Lowest tertile Intermediate tertile Highest tertile
Baseline (2007- 2011)
Follow-up (2012-
2016) Change
Baseline (2007- 2011)
Follow-up (2012-
2016) Change
Baseline (2007- 2011)
Follow-up (2012-
2016) Change
No. of countiesb 1041 1041 1041
Mean population across time period (adults aged 40-64 y)c
17 351 277 17 191486 −159 791 41 638 173 42 354 659 716 486 42 328 045 44 332 251 2 004 206
Baseline economic prosperity level, mean (SD)d
53.3 (28.0) 52.5 (28.4) 44.2 (29.4)
Economic prosperity marker, median (IQR)
Housing occupancy rate, %
90.1 (87.2 to 92.5)
88.5 (85.5 to 91.2)
−1.3 (−2.9 to −0.1)
90.6 (88.1 to 93.1)
90.4 (87.6 to 93.0)
−0.2 (−1.6 to 1.0)
89.9 (86.6 to 92.5)
90.7 (87.6 to 93.2)
0.8 (−0.5 to 2.2) Ratio of county median
household income to state median household income
0.9 (0.8 to 1.0)
0.8 (0.7 to 0.9)
−0.03 (−0.1 to 0.002)
0.9 (0.8 to 1.0)
0.9 (0.8 to 1.0)
−0.003 (−0.03 to 0.03)
0.9 (0.8 to 1.0)
0.9 (0.8 to 1.1)
0.03 (0.0 to 0.1)
Proportion of working adults aged 25-64 y, %e
70.1 (64.4 to 76.8)
67.8 (61.1 to 74.9)
−2.2 (−4.4 to −0.7)
71.2 (65.9 to 76.4)
70.9 (65.0 to 76.0)
−0.3 (−1.6 to 0.8)
69.7 (62.8 to 75.0)
70.9 (64.0 to 76.6)
1.1 (−0.1 to 2.6) Proportion of adults
with a high school education, %
85.0 (79.6 to 89.1)
86.6 (80.4 to 90.2)
1.1 (0.1 to 2.3)
85.8 (80.5 to 89.4)
87.7 (83.2 to 91.1)
1.8 (0.9 to 3.0)
84.3 (77.5 to 88.6)
87.1 (81.6 to 90.9)
2.8 (1.5 to 4.3)
Proportion of the population with income above the poverty threshold, %
85.2 (81.0 to 88.8)
83.3 (78.4 to 87.2)
−1.8 (−3.5 to −0.5)
85.3 (81.5 to 88.8)
84.6 (80.6 to 88.8)
−0.7 (−1.7 to 0.5)
84.2 (79.5 to 88.4)
85.3 (81.2 to 89.0)
0.6 (−0.6 to 2.3)
Percentage change in business establishmentsf
−4.9 (−7.8 to −1.4)
−1.7 (−5.0 to 1.1)
3.0 (−1.4 to 6.7)
−6.5 (−9.1 to −3.8)
0.6 (−2.4 to 3.6)
7.0 (3.3 to 10.4)
−7.4 (−10.9 to −4.2)
2.0 (−1.1 to 6.3)
10.0 (5.7 to 14.2)
Percentage change in employmentg
−3.7 (−8.5 to 1.4)
0.0 (−6.1 to 4.7)
3.3 (−5.3 to 10.4)
−6.3 (−11.1 to −1.7)
4.2 (−0.8 to 9.2)
10.8 (3.7 to 17.7)
−8.7 (−14.0 to −3.6)
8.1 (2.3 to 13.6)
17.3 (9.0 to 24.7) Abbreviation: IQR, interquartile range.
aChange in economic prosperity is the unweighted mean of the ranks for change in the 7 markers of economic prosperity between baseline (2007-2011) and follow-up (2012-2016). A higher mean rank indicates a greater relative increase or lower relative decrease in economic prosperity compared with lower-ranked counties. The mean rank for change in economic prosperity ranged from 5.4 to 43.8 for the lowest tertile, 43.8 to 56.0 for the intermediate tertile and 56.1 to 91.9 for the highest tertile.
bIncludes all US counties except for 20 counties with economic prosperity markers unavailable.
cBased on mean population during the baseline (2007-2011) or follow-up (2012-2016) periods.
dBaseline economic prosperity is the unweighted mean rank (scaled from 0 to 100) of 7 markers of economic prosperity in the baseline period (2007-2011).
eIndicates percentage of all 25- to 64-year-old residents in a county (including those not in the labor pool) who are employed. This differs from the employment rate, which is limited to individuals in the labor pool.
fRelative percentage change in the number of business establishments between the first and last years of the time period.
gRelative percentage change in the number of jobs between the first and last years of the time period.
mortality for non-Hispanic White individuals.21,22However, the current analysis suggests that for cardiovascular mortal-
ity, the association between economic prosperity and mortal- ity at the county level is significant for non-Hispanic Black Figure 2. Population-Weighted Mean Annual Age-Adjusted Mortality Rates (40- to 64-Year-Old Adults) by Tertile of Change in Economic Prosperity
500
400
300
200
100 600
0
Deaths per 100 000 individuals per year
Year Lowest tertile
Intermediate tertile Highest tertile
2010 2011 2012 2013 2014 2015 2016 2017
All causes
All cardiovascular disordersa Ischemic heart disease Stroke
Cause of death
Change in economic prosperity is the unweighted mean of the ranks for change in the 7 markers of economic prosperity between baseline (2007-2011) and follow-up (2012-2016). A higher mean rank indicates a greater relative increase or lower relative decrease in economic prosperity compared with lower-ranked counties. The mean rank for change in economic prosperity ranged from 5.4 to 43.8 for the lowest tertile (n = 1041), 43.8 to 56.0 for
the intermediate tertile (n = 1041), and 56.1 to 91.9 for the highest tertile (n = 1041). Counties with economic prosperity markers unavailable were not included.
aAll cardiovascular disorders include ischemic heart disease and stroke.
Table 3. Generalized Linear Mixed-Effects Model Estimates for Relative Additional Percentage Change in Mortality per Year With 10-Point Greater Change in Economic Prosperitya,b
Additional absolute change in mortality rate per year with 10-point greater change in economic prosperity, deaths per 100 000 individuals per year (95% CI)c
Additional annual percentage change in mortality per year with 10-point greater change in economic prosperity, % (95% CI)d Primary outcome: age-adjusted cardiovascular mortality rate
All residents aged 40-64 y 0.40 (0.22 to 0.58) fewer deaths −0.40 (−0.58 to −0.22)
Secondary outcomes
Age-adjusted cardiovascular mortality rate Individuals aged 40-64 y
Female 0.34 (0.16 to 0.51) fewer deaths −0.58 (−0.88 to −0.27)
Male 0.50 (0.18 to 0.82) fewer deaths −0.35 (−0.57 to −0.12)
Non-Hispanic Black 1.16 (0.53 to 1.80) fewer deaths −0.60 (−0.92 to −0.27)
Non-Hispanic White 0.49 (0.31 to 0.67) fewer deaths −0.53 (−0.72 to −0.33)
Hispanic (all races) 0.41 (0.06 to 0.76) fewer deaths −0.57 (−1.05 to −0.08)
Age-adjusted ischemic heart disease mortality ratee
All individuals aged 40-64 y 0.30 (0.17 to 0.43) fewer deaths −0.51 (−0.74 to −0.29)
Age-adjusted stroke disease mortality ratee
All individuals aged 40-64 y 0.04 (0.01 to 0.07) fewer deaths −0.26 (−0.47 to −0.04)
Age-adjusted all-cause mortality rate
All individuals aged 40-64 y 2.15 (1.62 to 2.68) fewer deaths −0.47 (−0.58 to −0.35)
aModel adjusted for baseline economic prosperity, year, percentage of residents living in rural areas in 2010, and the following time-varying covariates: percentage of residents who are female (except for sex subgroups), percentage of residents who are non-Hispanic Black (except for racial/ethnic subgroups), percentage of residents who are Hispanic (except for racial/ethnic subgroups), percentage of adult residents with diabetes, percentage of adult residents with obesity, primary care physicians per 100 000 residents, hospital beds per 100 000 residents, and percentage of residents with health insurance.
bChange in economic prosperity is the unweighted mean of the ranks for change in the 7 markers of economic prosperity between baseline (2007-2011) and follow-up (2012-2016). A 10-point greater change indicates 10 ranks
higher for mean rank for change in economic prosperity on a scale ranging from 5.4 to 91.9.
cIndicates the estimated absolute difference in the number of deaths per 100 000 individuals per year (based on population-weighted median mortality rate in 2010) between 2 counties with a difference of 10 ranks for mean rank for change in economic prosperity, holding all other variables constant.
dRegression estimate for interaction term between year and change in economic prosperity. Indicates the estimated difference in the annual percentage change in mortality between 2 counties with a difference of 10 ranks for mean rank for change in economic prosperity, holding all other variables constant.
eSubset of deaths from all cardiovascular disorders (primary outcome).
and Hispanic populations as well. This is plausible given the greater prevalence of cardiovascular risk factors, such as dia- betes and obesity, and on average higher rates of poverty among non-Hispanic Black and Hispanic populations in the US compared with the non-Hispanic White population.23-25
Limitations
This study has several limitations. First, due to the observa- tional design of the study, the associations noted cannot be concluded to be causal. Second, unmeasured confounding is likely, although several important demographic and health care–related variables were accounted for that have a plausible association with cardiovascular mortality. Third, because the economic prosperity markers were not avail- able for each individual year, it is not possible to rule out potential reverse causality, ie, an increase in cardiovascular mortality leading to worsening prosperity. However, it does allow for an analysis of the overall direction of change in prosperity in counties over the study period. Fourth, because all data are aggregated at the county level, infer-
ences at the individual level cannot be made. Fifth, the analysis relies on the recorded cause of death; it is possible that deaths from cardiovascular disease may be miscoded.26 However, when a broader definition (mortality from dis- eases of the circulatory system) was used, the results were concordant to the main analysis. Sixth, these results may not be generalizable beyond the study period of 2010 to 2017 and may not reflect the potential influence that eco- nomic distress related to the coronavirus disease 2019 pan- demic may have on cardiovascular mortality.
Conclusions
In this retrospective study of US county-level mortality data from 2010 to 2017, a relative increase in county-level eco- nomic prosperity was significantly associated with a small rela- tive decrease in cardiovascular mortality among middle- aged adults. Individual-level inferences are limited by the ecological nature of the study.
ARTICLE INFORMATION
Accepted for Publication: December 15, 2020.
Author Affiliations: Division of Cardiovascular Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia (Khatana, Nathan, Eberly); Penn Cardiovascular Outcomes, Quality, &
Evaluative Research Center, Perelman School of Medicine, University of Pennsylvania, Philadelphia (Khatana, Nathan, Dayoub, Eberly, Groeneveld);
The Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia (Khatana, Venkataramani, Nathan, Dayoub, Eberly, Mitra, Groeneveld); Department of Medical Ethics and Health Policy, Perelman School of Medicine, University of Pennsylvania, Philadelphia (Venkataramani); Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology, Beth Israel Deaconess Medical Center, Boston, Massachusetts (Kazi, Yeh); Division of Cardiology, Beth Israel Deaconess Medical Center, Boston, Massachusetts (Kazi, Yeh); Harvard Medical School, Boston, Massachusetts (Kazi, Yeh); Department of Biostatistics, Epidemiology, and Informatics, Perelman School of Medicine, University of Pennsylvania, Philadelphia (Mitra); Harvard T.H. Chan School of Public Health, Boston, Massachusetts (Subramanian); Division of General Internal Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia (Groeneveld); Center for Health Equity Research and Promotion, Michael J. Crescenz Veterans Affairs Medical Center, Philadelphia, Pennsylvania (Groeneveld).
Author Contributions: Dr Khatana had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Khatana, Nathan, Subramanian, Groeneveld.
Acquisition, analysis, or interpretation of data:
Khatana, Venkataramani, Dayoub, Eberly, Kazi, Yeh, Mitra.
Drafting of the manuscript: Khatana.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Khatana, Venkataramani, Dayoub, Mitra, Groeneveld.
Administrative, technical, or material support:
Eberly.
Supervision: Kazi, Subramanian, Groeneveld.
Conflict of Interest Disclosures: Dr Khatana reported receiving grants from the National Heart, Lung, and Blood Institute during the conduct of the study. Dr Venkataramani reported receiving grants from the Robert Wood Johnson Foundation, University of Wisconsin Center for Financial Security, and US Social Security Administration outside the submitted work. Dr Yeh reported receiving grants and consulting fees for unrelated work from Abbott Vascular and Medtronic.
No other disclosures were reported.
Funding/Support: Dr Khatana’s work is supported by grants from the National Heart, Lung, and Blood Institute (5T32HL007843-23, 1K23HL153772-01).
The Distressed Communities Index data were obtained under agreement from the Economic Innovation Group.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study;
collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The findings expressed in this article are solely those of the authors and not necessarily those of the Economic Innovation Group.
REFERENCES
1. Sidney S, Quesenberry CP Jr, Jaffe MG, et al.
Recent trends in cardiovascular mortality in the United States and public health goals. JAMA Cardiol.
2016;1(5):594-599. doi:10.1001/jamacardio.2016.1326 2. Chen Y, Freedman ND, Albert PS, et al.
Association of cardiovascular disease with premature mortality in the United States. JAMA Cardiol. 2019;4(12):1230-1238. doi:10.1001/
jamacardio.2019.3891
3. Curtin SC. Trends in cancer and heart disease death rates among adults aged 45-64:
United States, 1999-2017.Natl Vital Stat Rep. 2019;
68(5):1-9.
4. Wang SY, Tan ASL, Claggett B, et al. Longitudinal associations between income changes and incident cardiovascular disease: the Atherosclerosis Risk in Communities Study. JAMA Cardiol. 2019;4(12):
1203-1212. doi:10.1001/jamacardio.2019.3788 5. Semega J, Kollar M, Creamer J, Mohanty A.
Current Population Reports, Report No.
P60-266(RV): Income and Poverty in the United States: 2018. US Census Bureau; 2019.
6. Havranek EP, Mujahid MS, Barr DA, et al;
American Heart Association Council on Quality of Care and Outcomes Research, Council on Epidemiology and Prevention, Council on Cardiovascular and Stroke Nursing, Council on Lifestyle and Cardiometabolic Health, and Stroke Council. Social determinants of risk and outcomes for cardiovascular disease: a scientific statement from the American Heart Association. Circulation.
2015;132(9):873-898. doi:10.1161/CIR.
0000000000000228
7. Lòpez-De Fede A, Stewart JE, Hardin JW, Mayfield-Smith K. Comparison of small-area deprivation measures as predictors of chronic disease burden in a low-income population. Int J Equity Health. 2016;15:89. doi:10.1186/s12939-016- 0378-9
8. Durfey SNM, Kind AJH, Buckingham WR, DuGoff EH, Trivedi AN. Neighborhood disadvantage and chronic disease management. Health Serv Res.
2019;54(suppl 1):206-216. doi:10.1111/1475-6773.
13092
9. National Center for Health Statistics. Detailed Mortality—All Counties: 2010-2017, as Compiled From Data Provided by the 57 Vital Statistics Jurisdictions From the Vital Statistics Cooperative Program. National Center for Health Statistics; 2018.
10. Economic Innovation Group. Distressed Communities Index: methodology. Accessed March 28, 2020.https://eig.org/dci/methodology 11. Foreman KJ, Naghavi M, Ezzati M. Improving the usefulness of US mortality data: new methods for
reclassification of underlying cause of death. Popul Health Metr. 2016;14:14. doi:10.1186/s12963-016- 0082-4
12. Arias E, Heron M, Hakes J; National Center for Health Statistics; US Census Bureau. The validity of race and Hispanic-origin reporting on death certificates in the United States: an update.Vital Health Stat 2. 2016;(172):1-21.
13. Lee S, Colditz GA, Berkman LF, Kawachi I.
Prospective study of job insecurity and coronary heart disease in US women. Ann Epidemiol. 2004;
14(1):24-30. doi:10.1016/S1047-2797(03)00074-7 14. Gallo WT, Bradley EH, Falba TA, et al.
Involuntary job loss as a risk factor for subsequent myocardial infarction and stroke: findings from the Health and Retirement Survey. Am J Ind Med.
2004;45(5):408-416. doi:10.1002/ajim.20004 15. Kawachi I, Kennedy BP. Health and social cohesion: why care about income inequality? BMJ.
1997;314(7086):1037-1040. doi:10.1136/bmj.314.
7086.1037
16. Shanahan L, Hill SN, Gaydosh LM, et al. Does despair really kill? a roadmap for an evidence-based answer. Am J Public Health. 2019;109(6):854-858.
doi:10.2105/AJPH.2019.305016
17. Tawakol A, Osborne MT, Wang Y, et al.
Stress-associated neurobiological pathway linking socioeconomic disparities to cardiovascular disease. J Am Coll Cardiol. 2019;73(25):3243-3255.
doi:10.1016/j.jacc.2019.04.042
18. Vaughan AS, Ritchey MD, Hannan J, Kramer MR, Casper M. Widespread recent increases in county-level heart disease mortality across age groups. Ann Epidemiol. 2017;27(12):796-800. doi:
10.1016/j.annepidem.2017.10.012 19. Cross SH, Mehra MR, Bhatt DL, et al.
Rural-urban differences in cardiovascular mortality in the US, 1999-2017. JAMA. 2020;323(18):1852-1854.
doi:10.1001/jama.2020.2047
20. United States Department of Agriculture Economic Research Service. Rural employment and unemployment. Updated September 23, 2019.
Accessed April 14, 2020.https://www.ers.usda.gov/
topics/rural-economy-population/employment- education/rural-employment-and-unemployment/
21. Stein EM, Gennuso KP, Ugboaja DC, Remington PL. The epidemic of despair among White Americans: trends in the leading causes of premature death, 1999-2015. Am J Public Health.
2017;107(10):1541-1547. doi:10.2105/AJPH.2017.
303941
22. Case A, Deaton A. Rising morbidity and mortality in midlife among white non-Hispanic Americans in the 21st century. Proc Natl Acad Sci U S A. 2015;112(49):15078-15083. doi:10.1073/pnas.
1518393112
23. Centers for Disease Control and Prevention.
National Diabetes Statistics Report, 2020. Centers for Disease Control and Prevention; 2020.
24. Arroyo-Johnson C, Mincey KD. Obesity epidemiology worldwide. Gastroenterol Clin North Am. 2016;45(4):571-579. doi:10.1016/j.gtc.2016.07.
012
25. Kaiser Family Foundation. State health facts:
poverty race by race/ethnicity. Accessed April 15, 2020.https://www.kff.org/other/state-indicator/
poverty-rate-by-raceethnicity/
26. Pagidipati NJ, Gaziano TA. Estimating deaths from cardiovascular disease: a review of global methodologies of mortality measurement.
Circulation. 2013;127(6):749-756. doi:10.1161/
CIRCULATIONAHA.112.128413
